Visualising hierarchical multiple regression

Hi everyone,

Apologises if this is a stupid question, but I have been trying to find an answer all day and am not really getting anywhere. I have performed hierarchical multiple regression in SPSS and am unsure if there is any way to visually present the tables (in graphical form) so it looks more aesthetically pleasing? I don't think this is possible, as the papers I've looked at don't seem to include anything in terms of graphical representations and stick to the tables, but I thought I would ask here just in case. I am able to present the partial plots but this only works for one variable at a time (in relation to the dependent variable) - there doesn't seem to be a way to present more than one variable at a time?

Again, sorry if this is a silly question, I'm quite new to statistics!
My data is looking at mathematics and what factors significantly increase or decrease interest in the subject, including age, gender, self-efficacy, that kind of thing.
I have produced an 'interest in STEM' variable produced from a factor analysis of 8 questionnaire questions, 'How would you rate your interest in [subject]' on a scale of 0-5. There were eight subjects, and using factor analysis, five subjects (including maths, chemistry etc) were factored into an 'interest in STEM' variable, which is my dependent variable, and the remaining three into a 'interest in humanities' variable, which is one of my factors in the multiple regression. So interest in STEM is a factored variable produced from this factor analysis.
I have a few binary variables in the analysis though - 'does your father have a degree' (yes or no), the same for 'does your mother have a degree', and then 'I have female role models that are in STEM fields'.


Less is more. Stay pure. Stay poor.
How is your model structured as in the following would be used in R for a random intercept and effects model:

Interest ~ 1 + mother + father + gender ( 1 + ?variable | cluster_ID)

What is your sample size and number of groups in the clusters?
When you say hierarchical multiple regression do you mean:

(1) You've added independent variables into the final model in stages (e.g. model with 1 IV, model with 1IV and another, etc, etc)


(2) A regression model with random effects (which can be viewed as a Multilevel Model)

I believe you might mean the former, but just want to double check.
No problem.

It sounds like you might be from a behavioral science background (from where I also hail) and I know the terminology can be different. For example if you were to say mixed model, you might mean one between factor and one within factor, where as in other areas, people might interpret this as having both fixed and random effects.

Anyway, from what I recall from doing these analyses, we just looked at the increase / decrease in the adjusted R-Square after running each model. Framed this way, you can have a line or bar plot representing the R-square value at each stage of the model (e.g. adjusted R square value on the y axis going from 0 to 1, and on the x-axis discrete points, each representing a new model)

Try googling Variable Importance Plot to get a general idea

Last edited:
I am a psychology student, so yes! Yes, I think that sounds like the best way, considering the R square change is the most useful element, so plotting that makes sense - thank you so much for your help (both of you!)